Tech

You Can Choose New 'Thinking Levels' for Gemini

By Alex • May 31, 2026
You Can Choose New 'Thinking Levels' for Gemini

You Can Choose New 'Thinking Levels' for Gemini

Let's be real: AI terminology is a minefield. We talk about "intelligence," "understanding," and yes, even "thinking." In the fast-paced world of large language models, "thinking" usually refers to a specific type of process where an AI breaks down a request into multiple steps, giving it more time to "think through" a complex problem before spitting out an answer. These dedicated "thinking models" have been around, but if you're a Gemini user, you've just gained a new layer of control over how your AI assistant "thinks"—even when you're not using those specialized, slower models.

Google is rolling out new "Thinking Levels" within its Gemini app for both iOS and Android, adding a crucial layer of nuance to your interactions. This isn't the existing, resource-intensive "Thinking" model under Gemini 3. Instead, these new "levels" let you fine-tune the processing depth when you're using faster options like Gemini 3 Flash (branded as "Fast") or Gemini 3.1 Pro.

Boost Your Brainy Bot Power

Imagine you're using Gemini's "Fast" option. It's speedy, great for quick queries, but sometimes you need a bit more depth without waiting for the full "Thinking" model experience. That's where these new levels come in.

Here's how to tap into this new feature:

  1. Open the Gemini App: Launch Gemini on your iPhone or Android device.
  2. Start a New Chat: Or continue an existing one.
  3. Check Your Model: Make sure you're using either "Fast" (Gemini 3 Flash) or "Gemini 3.1 Pro."
  4. Find "Thinking Level": Look for a new "Thinking level" option at the bottom of the model selector.
  5. Choose Your Depth: You'll see two options:
    • Standard: Google says this is optimal for most everyday problems. It's the default, balanced approach.
    • Extended: This setting gives Gemini more processing time for truly "complex topics." Expect it to take a little longer, but you should see a noticeable improvement in the quality and thoroughness of the response.

When to Go 'Extended'

Think of "Extended" as giving your Gemini assistant a bit more quiet time to really noodle on your prompt. If you're tackling:

  • In-depth research questions: Asking for nuanced comparisons or detailed explanations.
  • Creative writing prompts: Where a generic answer won't cut it.
  • Complex problem-solving: Requiring multi-step reasoning.
  • Code generation: Especially for intricate functions or debugging scenarios.

Switch to "Extended." It's a sweet spot between lightning-fast but potentially superficial, and super-deep but agonizingly slow. This makes your current models more versatile, letting you optimize for speed or quality on the fly.

The best part? This feature appears to be rolling out to all users, not just Google AI Plus subscribers. So, keep an eye on your Gemini app—you'll likely see these new "Thinking Levels" popping up soon. With Google I/O just around the corner, expect even more exciting AI developments heading your way!